Optimization and classification of regularized mixtures of scaled Gaussian distributions, a.k.a regularized compound Gaussian distributions

This repository hosts Python code for the numerical experiments of the the associated arXiv paper and relies on the library pyCovariance.

Installation

The script install.sh creates a conda environment with everything needed to run the examples of this repo and installs the package:

./install.sh

Check

To check the installation, activate the created conda environment optim_compound and run the unit tests:

conda activate optim_compound
nose2 -v --with-coverage

Run numerical experiments

To run experiments, run the scripts from the different folders center_of_mass/, classification/, estimation/ e.g.

python estimation/speed_comparison.py

Cite

If you use this code please cite:

@misc{collas22MSG,
      title = {Riemannian optimization for non-centered mixture of scaled Gaussian distributions},
      author = {Collas, Antoine and Breloy, Arnaud and Ren, Chengfang and Ginolhac, Guillaume and Ovarlez, Jean-Philippe},
      year = {2022},
      url = {https://arxiv.org/abs/2209.03315}
}